By directing analysis toward the power and axis of astigmatism, this study reveals several associations between cylindrical refractive errors and parameters related to the subjects' overall refractive status, to their individual and familial characteristics, and to available potential early life environmental exposures. It has long been appreciated that incorporating astigmatism complicates the classification of the refractive status of individual subjects.
11,22 In the common spherical equivalent notation (spherical power + half of the cylinder power) or a more recently proposed transformation of refraction values through Fourier analysis,
23 as examples, the spherical component of refraction is influenced arithmetically by the power of the cylinder. Since both the spherical power and cylindrical power occur in the same eye, they are not strictly independent either biologically or statistically. Whether the developmental mechanisms causing spherical and cylindrical ametropia are actually independent, however, is not known. In an effort to separate at least mathematically the spherical and cylindrical powers, Guggenheim and Farbrother
22 have proposed the LDE (least deviation from emmetropia) classification scheme incorporated here.
Applying this LDE classification scheme to a large population of Israeli military conscripts revealed complex relationships between the spherical power and the cylindrical axis and power (
Figs. 1 2–
3). With LDE refractions deviating from emmetropia in either the myopic or hyperopic direction, WTR astigmatic axes became more prevalent with larger ametropia. For ATR axis, the opposite relationship held, with the largest proportion of ATR astigmatism occurring near emmetropia and with a decreasing prevalence as LDE increased in magnitude. In addition, cylinder axis was affected by cylinder power: the prevalence of WTR astigmatism increases with increasing cylinder power; that of ATR astigmatism decreases with increasing cylinder power. These complex relations between the prevalence of WTR and ATR astigmatism with LDE and cylinder powers conform to a prior report in which the same approach was used to classify refraction.
10 In addition, our large sample size permitted assessment of OBL astigmatism. Although the magnitude of the effects were comparatively small, the prevalence of OBL astigmatism increased with increasing ametropia in either direction similar to WTR astigmatism (
Fig. 1); and the prevalence of OBL astigmatism decreased with increasing cylinder size, similar to ATR astigmatism (
Fig. 2).
Perhaps, the LDE notation used in our study and in a prior report
10 facilitated revealing these relationships between the cylinder direction and both spherical and cylindrical powers. These results suggest a hypothesis that, at least for the teenagers and young adults whom we studied, risk factors for WTR or OBL astigmatism and spherical ametropias may be similar, but the risk factors for ATR astigmatism may differ from those for underlying spherical ametropia (
Fig. 1). Conceivably, risk factors for ATR astigmatism may even be protective against ametropia. The reducing prevalence of ATR axis and OBL axis with increasing cylinder power (
Fig. 2) also raises the possibility of different risk factors for the two parameters of axis orientation and astigmatism power. However, the present study was a cross-sectional survey that assessed only astigmatic subjects and not a prospective study, and validating such causative hypotheses will require more definitive methods in the future.
Both mechanical and nonmechanical causes have been invoked as a basis for astigmatism. Most commonly discussed are mechanical hypotheses, such as the exertion of tension from extraocular muscles or eyelid weight on the cornea.
7 Eyelid anatomy, orbital anatomy, and related external pressures on the globe have been associated with astigmatism axis and corneal shape.
24,25 Transient effects of eye position on corneal astigmatism are generally assumed to result from an altered relationship of the eyelid and cornea,
26 but the action of the extraocular muscles cannot strictly be excluded in such evaluations. Eyelid surgery can have modest effects on corneal astigmatism,
27 but the effect may be transient in many patients.
28 On the other hand, the lack of correlation between corneal toricity and direct measurements of eyelid tension
29 and the marked differences in cylinder orientation between Chinese and Caucasian infants
30 suggest that other nonmechanical factors may be at least partly responsible for astigmatism. The balancing of corneal and internal sources of astigmatism to reduce the toricity in individual eyes
31 also suggests a role for nonmechanical biological processes in generating astigmatism, as now seemingly underlies refractive errors in general.
12
Although they were not as pronounced as in other studies reporting the ethnic differences in the prevalence of astigmatism,
6–8,31 we did detect modest differences in the proportion of WTR, ATR, or OBL axes among astigmatic conscripts whose families originated from different geographic regions. Israeli and Eastern ethnic origin were associated with more WTR axis compared with Western origin. Ethnic differences in astigmatism axis conform with a prior study from Australia
31 that found higher proportions of WTR and less ATR astigmatism axis in children of non–European Caucasian ethnicity compared to children of Asian ethnicity. Whether the effect of Israeli conscript origin on astigmatism prevalence are explainable as differences in genetic admixing in prior generations, by differences in living environments in Israel
6,32 or perhaps by ethnic differences in orbital and lid structure cannot be assessed from the current data.
Indices of body stature have repeatedly been studied in association with refractive errors, most commonly as spherical ametropias; but results vary between studies, and the mechanisms relating body stature and refraction are poorly understood.
33–35 In the present Israeli conscript data, WTR astigmatism increased with increasing BMI (
Fig. 4), but the prevalence of both ATR and OBL astigmatism decreased with increasing BMI. Perhaps, the higher prevalence of WTR in obese compared with nonobese young adults may be explainable by a rise in eyelid tension on the globe with increasing body weight and a resultant steepening of the corneal vertical meridian and WTR astigmatism.
Higher intelligence, as measured by standard tests, has long been associated with increased myopia prevalence.
36,37 Similarly, in the current data, higher intelligence scores were associated with greater myopia. In addition, we found that higher intelligence scores were associated with lower magnitude of cylinder power, higher prevalence of ATR axis orientation, and lower prevalence of both WTR and OBL axis orientations. Of note, the associations of intelligence with astigmatism parameters were dissociated from the associations of intelligence with the spherical component of refraction based on the following considerations. Intelligence scores increased with increasing myopia, with decreasing cylinder power, and with higher prevalence of ATR axis orientations; but increasing myopic LDE was associated with less ATR axis (
Fig. 1), precisely the opposite direction anticipated if the associations with intelligence scores were consistent across all refractive parameters.
Although the effect was comparatively small, photoperiod length at the time of birth also showed complex associations with cylinder axis. Among the astigmatic subjects, the prevalence of WTR axes increased with birthdates in the months with longer photoperiods (
Table 1). In this population, higher degrees of myopia had been associated with birthdates in months with longer photoperiod.
15 Because WTR axes also increased with increasing hyperopic LDE (
Fig. 1), the photoperiod relationships suggest differences in the risk factors for the astigmatic and spherical components of refraction. The association between astigmatism and season of birth or perinatal photoperiod, identified in this study, further emphasizes the potential influence of perinatal or neonatal environmental parameters on refractive development (Fotedar R, et al.
IOVS 2008;49:ARVO E-Abstract 3142).
13,14 As recently discussed,
15,17 it is not known whether the birth date and photoperiod associations with refraction are causal or instead reveal the influences of different environmental parameters. In fact, available data do not distinguish whether the primary influence acts directly on the infant after birth or indirectly through the mother during pregnancy or breastfeeding. For instance, photoperiod may associate with seasonal differences in ambient temperature or diet that have not been investigated in relation to subsequent refractive development. Despite much research, the role of environmental parameters in refractive development remains poorly understood. That perinatal or neonatal environmental parameters might exert long-term influences on refraction is an evolving, novel, and yet controversial notion. Clearly, more research is needed to further advance the mechanistic understanding in this field.
In our study, being female was associated with slightly more WTR axis compared with being male, an association that was statistically significant in multivariate but not in univariate analysis. Although the relations of astigmatism and sex have varied in prior studies, our results are consistent with those in a recent report
31 of a higher percentage of WTR and lower percentage of OBL astigmatism in females. In another recent report in which both astigmatism and features of eyelid anatomy were examined, astigmatism axis correlated with the orientation of the palpebral fissure slant; as the females had more of an upward palpebral fissure slant in the Caucasian and African-American subjects studied, the sex difference in cylinder axis orientation may relate to such sex-related differences in the orientation of the lid fissures.
38 Nonmechanical explanations for the sex effect are also possible, although, as was observed in a small study, a correlation of lower blood estrogen levels and increased horizontal corneal curvature in postmenopausal women.
39
Several methodological considerations are pertinent to the associations found in the present study. First is the use of the recent LDE approach for refraction classifications. Many notations for refractive error are available; but for most, the astigmatism power affects and thus confounds the spherical classification. Although not without its own peculiarities, the LDE notation at least minimizes this sort of potential classification error. Second, we included data from right eyes only. Although a common approach in many studies of refraction to avoid statistical confounding by nonindependence of the two eyes, including only right eyes precludes addressing either direct or mirror symmetry of astigmatism between the eyes of individual subjects.
40
Total astigmatism as measured in this study results from the combination of corneal and internal astigmatism, the latter chiefly deriving from the ocular lens.
7 In population averages, internal astigmatism tends to be small in power and ATR in orientation, but with some individual variability.
41,42 Because keratometry readings are not included in the conscription eye examination, we cannot reliably distinguish the corneal versus internal origins of overall ocular astigmatism in our subjects. How the parameters that we studied relate to corneal and internal astigmatism and whether the association patterns are similar or distinct will certainly be informative questions for future research.
In the present analysis, we were not able to exclude data from subjects with ocular or systemic diseases that have might have secondarily affected astigmatism. However, in a similar (and partially parallel) sample of refraction data from the Israeli Defense Forces, the prevalence of keratoconus among the astigmatic population was estimated to be <0.4:1000 (Mandel Y, unpublished data, 2009). Moreover, subjects with severe systemic diseases are usually excluded from military service and would not routinely be refracted or have been included in the current data set. Thus, we believe that the prevalence of ocular or systemic conditions potentially influencing astigmatism is small and unlikely to be inducing a significant bias.
The present analysis necessarily depended on data available only from the preconscription examination; but the large sample size, comparatively homogeneous population, and narrow age range are likely to minimize potential confounding in the findings. Two striking conclusions emerged from the summary provided in
Table 4. First, the associations with ATR, WTR, and OBL astigmatism axes were neither parallel as a whole nor parallel between any two sets of axis orientations. Second, the directions of the associations with astigmatism axis and astigmatism power did not parallel previously reported associations for spherical ametropias. For instance, both WTR astigmatism and myopia prevalence increased with increasing birth month photoperiod in this population
15 ; but WTR astigmatism prevalence decreased with increasing intelligence, just the opposite association long found for myopia and intelligence.
36,37 Although refractive development can continue beyond the beginning of the third decade, the refractions of military aged subjects should represent relatively mature ocular development. Many of the epidemiologic parameters studied here have long been incorporated in refractive research. Presumably, such readily identifiable clinical parameters are surrogate markers, however imperfect, for the molecular and cellular mechanisms that govern the growth and form of the eye. To the extent that associations with these clinical parameters may reveal the underlying biological mechanism of refractive errors, the complexity of the present findings suggests both that dissimilar mechanisms are responsible for the different forms of astigmatism and that mechanisms governing development of astigmatism likely diverge from those governing the spherical component of refraction.
Supported by The Paul and Evanina Bell Mackall Foundation Trust (RAS) and Research to Prevent Blindness (RAS).